Modeling a Multi-Objective Hub Location Problem with Multiple Allocation under Uncertainty
A hub location allocation problem is one of the applicable and discussable issues in recent years. The reason is significance of the cost reducing in problems (e.g., good transportation, telecommunication and passenger fleets) related to the network involving hub and spoke nodes. This thesis considers one specific type of a hub problem, namely covering problem. Related to this issue, various mathematical models are planned and an appropriate solution approach for each model is presented.Associated with the first and second mathematical models, one objective function and for the third mathematical model two objective functions and for the fourth and fifth models three objective functions are considered. The emphasis is on the final mathematical models with two and three. In the first model, in addition to the constraints related to the hub covering problems, various features (e.g., production facilities located in hubs being active during time horizons) are considered. Depending on the related radius, these facilities serve to the client nodes. In the second model, transportation vehicles are also considered. In the third model, a new objective function aiming to minimize the transportation time is accounted for. In the fourth model, issues (e.g., queuing and scheduling involving tardiness and earliness times) are taken into account. Additionally, minimizing emitted greenhouse gas by transporter vehicles navigating in roads and paths is taken into account. Related to this issue, numerous transportation modes are considered. Depending on the transportation mode, related transporter vehicles are considered. In the fifth mathematical model, a tri-objective mathematical model is considered, in which the costs of transportation, traffic and hub installation are minimized. To solve the presented mathematical models, novel approaches concentrating on the uncertainty matters are used. This relates to some parameters depending on the being strategic and being possible as indefinite ones in order to come close to reality. Therefore, indefinite parameters are considered. In such circumstance that data and experiences are predefined, stochastic and robust optimization approaches are considered.
In other problems, multi-objective mathematical model are considered. Related to the final mathematical model, five meta-heuristic algorithms are presented. In large-scale problems, the computational results depending on the important and affecting parameters are analyzed and presented. Related to the first solution approach, a stochastic mathematical model yields better results. Additionally, by increasing the parameter values both stochastic and deterministic problems present increasing trend. Generally, Related to the robust mathematical model by increasing the uncertainty level, the objective function value is increased. In addition, related to the third and fourth mathematical models, the ideal values of objective functions are computed and reported. At last, in the fifth mathematical model, the results obtained by five meta-heuristic algorithms are compared with each other using the related metrics in order to analyze the Pareto fronts in large-scale problems.
KEYWORDS : Hub location-allocation, Uncertainty, Queuing Theory, Scheduling, Production Facilities, Pareto Front.